FACES Manual
FACES Manual
FACES Manual
A Users Guide
Ann M. Kring
Department of Psychology
3210 Tolman Hall
University of California, Berkeley
Berkeley, CA 94720
Denise Sloan
Department of Psychology
Weiss Hall
Temple University
Philadelphia, PA 19122
Facial expression of emotion is of great interest to many researchers. It has been studied
in connection with subjective emotional experience, physiological arousal, and communication
to name but a few areas. Interest in facial expression has a rich history dating back to the mid
19th century (Piderit, 1858, 1888; Gratiolet, 1865). Perhaps the most influential of these early
theorists was Charles Darwin. In his book, The Expression of the Emotions in Man and Animals,
Darwin (1896) argued that facial expressions were universal and innate characteristics. In the
infancy of the science of psychology, William James hypothesized that facial expression
played a causative role in the experience of emotion. In fact, according to James, changes in the
facial musculature comprised a large portion of emotional state. James' ideas about emotion set
forth a tradition of scholarly debate about the role of facial expression in emotion that
continues today. In the early 1960's, Tomkins (1962, 1963) proposed what has become known
as the facial feedback hypothesis. Stated succinctly, facial feedback theory holds that facial
expression provides feedback which in turn produces the emotion. A tradition of research
investigating the facial feedback hypothesis (see Adelmann & Zajonc, 1989 for a review) has
ensued, but several unanswered questions remain regarding the mechanisms and functions of
facial expressions of emotion.
Ekman and Friesen (1976, 1978) were pioneers in the development of measurement
systems for facial expression. Their system, known as the Facial Action Coding System or
FACS, was developed based on a discrete emotions theoretical perspective and is designed to
measure specific facial muscle movements. A second system, EMFACS, is an abbreviated
version of FACS that assesses only those muscle movements believed to be associated with
emotional expressions. In developing these systems, Ekman importantly distinguishes between
two different types of judgments: those made about behavior (measuring sign vehicles) and
those that make inferences about behavior (message judgments). Ekman has argued that
measuring specific facial muscle movements (referred to as action units in FACS) is a
descriptive analysis of behavior, whereas measuring facial expressions such as anger or
happiness is an inferential process whereby assumptions about underlying psychological states
are made. It is important to point out, as Ekman does, that any observational system requires
inferences about that which is being measured. Other available systems have been designed to
measure either specific aspects of facial behavior (e.g., Ermiane & Gergerian, 1978; Izard,
1979; see Ekman (1982) for a selective review) or more generally defined facial expressions
(e.g., Notarious & Levenson, 1979).
The primary reason for developing a new system was based on the perceived need for a
facial coding that is theoretically aligned with a dimensional model of emotion. Several
researchers have argued that affective expression consists of two broad dimensions: valence
and arousal (e.g., Russell, 1980; Schlosberg, 1952). Similarly, researchers have argued that
emotional experience variance is also best captured by two dimensions (e.g., Larsen & Diener,
2
1992; Watson, Clark, & Tellegen, 1988). Most currently available coding systems of facial
behavior are based on discrete emotion theory and are designed to measure a number of
specific or basic emotions. Although some might argue that these systems can be considered
"dimensional" to the extent that discrete categories can be combined to form dimensions, this
approach is inconsistent with the empirical literature upon which dimensional models of
emotion have developed. The Facial Expression Coding System (FACES) was designed as a
dimensional measure of facial behavior.
Second, while the Ekman and Friesen systems have been the pacesetters for studying
facial expression of emotion, they are not without cost. It takes a great deal of time to train
coders to use the system accurately and reliably (Ekman (1982) estimated approximately 100
hours were needed for training). Additionally, coding time can be quite extensive and as a
result, often only small segments of participants' facial behavior are coded with FACS.
EMFACS is somewhat more economical in that coders are not required to detect each muscle
change but rather decide if a group of changes presumed to be associated with particular
emotions have occurred. Being restricted to examine small portions of a participants data,
although useful if researchers are interested in identifying specific responses to specific stimuli,
can also be problematic. First, examining small segments may obscure an examination of the
natural unfolding of expressive behavior over time. Second, selection of these segments most
often requires a priori decisions about which segment is likely to produce the most expressive
behavior. Selecting segments which maximize the likelihood of expressive behavior for all
participants can be quite difficult.
An Overview of FACES
The Facial Expression Coding System (FACES) was developed as a less time
consuming alternative to measuring facial expression that is aligned with dimensional models
of emotion. The system provides information about the frequency, intensity, valence, and
duration of facial expressions. The selection of the variables included in the system was based
on theory and previous empirical studies. Adopting the descriptive style of Ekman and similar
to the work of Notarious and Levenson (1979), an expression is defined as any change in the
face from a neutral display (i.e., no expression) to a non-neutral display and back to a neutral
display. When this activity occurs, a frequency count of expressions is initiated. Next, coders
rate the valence (positive or negative) and the intensity of each expression detected. Notice that
this is quite different from assigning an emotion term to each expression. While FACES
requires coders to decide whether an expression is positive or negative, it does not require the
application of specific labels. There is support in the literature for this approach, often referred
to as the cultural informants approach (Gottman & Levenson, 1985). That is, judgments about
emotion, in this case whether an expression is positive or negative, are made by persons who
are considered to be familiar with emotion in a particular culture. In addition to valence and
intensity, coders also record the duration of the expression. Finally, a global expressiveness
rating for each segment is made, and judgements about specific emotions expressed throughout
the segment can also be obtained.
3
How to Use FACES
FACES was initially developed to measure facial expressions in response to five minute
film clips. The system can be adapted to other applications, however, and attempts to represent
the broad applicability of the system are made throughout the manual. Generally speaking, the
system allows for the examination of a participant's entire record of expressive behavior. When
we videotaped participants viewing emotional films, the soundtrack from the movie was not
included on their videotapes. Thus, coders only viewed participants facial reactions to the
films. We have typically had two raters coding each participant. As will be discussed below,
reliability for FACES has routinely been very high.
Detecting an Expression
While viewing a participant's record, an expression is detected if the face changes from
a neutral display to a non-neutral display and then back to a neutral display. It is important to
note, however, that a facial display may not always return to the original neutral display but
may instead return to a display that, although neutral, does not exactly resemble the prior
neutral expression. Additionally, if after a participant displays a shift from a neutral to non-
neutral display and, instead of returning to a neutral display, shows a clear change in affective
expression, this change is counted as an additional expression. For example, if while smiling, a
participant then raises his or her eyebrows and stops smiling, indicating more of a surprised
look, two expressions will be coded.
Duration
Once an expression has been detected, the duration is measured. For convenience, a
time-mark in seconds should be included on participants' videotape. The duration measurement
should start as soon as the participant changes from a neutral to non-neutral display. This time
should be recorded on a coding form (sample coding forms are presented in the Appendix).
The duration measurement should stop as soon as the participant changes back from a
non-neutral to neutral display, and the time should be recorded on the coding form. The
duration in seconds can then be calculated by subtracting the beginning time from the end time
and then recorded on the coding form. Mean duration can be calculated by dividing the total
duration by the frequency of expressions. Typically this is done separately for positive and
negative expressions.
Valence
Next, the coder must decide whether the expression was positive or negative and make
the appropriate notation on the coding form. If there is doubt as to whether the expression is
positive or negative, a comprehensive list of affect descriptors is presented in the Appendix.
Extensive research (Russell, 1980; Watson & Tellegen, 1985) has established these descriptors
as either positive or negative. They are provided simply as a guide for coders in determining
the valence of an expression. Coders are not asked to supply a descriptor for each expression
detected.
4
Intensity
Intensity ratings for an individual expression range from one to four (1=low,
2=medium, 3=high, 4=very high). The low rating is given for those expressions that are mild,
such as a smile where a participant slightly raises the corners of his/her mouth but does not
show the teeth, and very little movement around the eyes occurs. The medium rating is given
for those expressions where a participant's expression is more moderate than mild in intensity,
such as a smile bordering on a laugh, with the eyebrows slightly raised and the lips apart,
exposing the teeth. The high rating is given for an expression that involves most, if not all, of
the face, such as laughing with an open mouth and raising the eyebrows and cheeks. The very
high rating is reserved for those expressions that are very intense. An example of such an
expression is one where a participant is undeniably laughing, with the mouth completely open
with the eyebrows and cheeks substantially raised.
When film clips are the stimuli, we have found it useful to provide summary
information at the end of each film clip. Specifically, two, subjective global ratings are taken:
judgments about the specific emotion(s) being expressed and a judgment about the overall level
of expressiveness. Additionally, summary information is calculated for the frequency, intensity,
and duration measures for both positive and negative expressions. two sample summary sheets
are found in the Appendix.
Although not a primary focus of the system, we have used two different rating schemes
to assess more specific judgements of individual emotions: a forced choice rating and Likert
format ratings. Using the forced choice method to determine the predominant emotion
expressed, the coder should look over the coded expressions for the entire segment to obtain an
assessment of whether the participant was expressing predominantly positive or negative
emotions. Then, the coder is required to choose one of seven emotions on the summary form
(happiness, sadness, disgust/fear, interest, neutral/indifferent, surprise, or anger). These were
chosen as manipulation checks for the emotional film clips and can certainly be modified for
different applications. This can be a difficult item to code. For example, a participant who was
expressive during the segment can still obtain a global rating of neutral/indifferent if the
expressions were all low in intensity and short in duration.
Using Likert scales, coders are required to rate, using a six point scale (1 = not at all to
6 = very much), the degree to which each specific emotion (e.g., happiness, sadness,
amusement, fear, disgust, anger, interest) was expressed during the segment.
Level of Expressiveness
This rating is the coder's global assessment of expressiveness during a segment. Before
making this assessment, the coder should look at all the individual expression ratings during
5
the segment. That is, this rating requires consideration of individual ratings of valence,
duration, and intensity. The global rating of expressiveness ranges from one to five (1=low,
2=fairly low, 3=moderate, 4=fairly high, 5=high). A low rating would be given to a participant
who had none or few expressions all of which were short and low in intensity. In contrast, a
high rating would be assigned to a participant who had many highly intense and longer
expressions.
Summary Measures
Experience with the coding system tells us that there are a number of things that can be
problematic for coders if they are not discussed ahead of time. Below, we provide a list of the
most common problems. This list is necessarily tied to our application and thus may not be
applicable in other studies or settings. These suggestions are offered as guides, not absolute
solutions, for coders.
A coder may sometimes mistake a change of body position for a change from a neutral
to non-neutral facial display. Coders must take special care to ensure that the face changes in
addition to the body posture shift in order to record that an expression has occurred.
The coder should not code any expressions if the participant does not appear to be
paying attention to the stimuli. Although this can be difficult to determine, if the participant is
looking down or away from the stimulus that is being presented, it is likely that he or she is not
attending. We have also employed separate ratings of attention in order to assess this more
systematically. Depending on the application and participant population, this may be advisable.
6
Hand covering part of face
If a participant's hand is covering part of the face, the coder may unfortunately need to
rely on the other parts of the face to detect the occurrence of facial expressions. For example, if
the participant is covering the mouth area, the coder will need to pay special attention to the
eye, nose, forehead, and cheek areas to code expressions.
Eye glasses
If the participant is wearing eye glasses, the coder may find it difficult to examine the
participant's eyes during facial expressions. In this situation, the coder is encouraged to
examine as best as possible eye movements (e.g., eyebrows raised above the eyeglass frame) as
well as other areas of the face when determining whether or not an expression has occurred.
Contact lenses
If the participant is a contact lens wearer, chances are good that he or she may have eye
movements related to the lenses and not to facial expressions per se. If possible, determine
ahead of time if the participant wears and/or experiences any problems with contact lenses. If
repetitive movements (e.g., blinking) occur that do not appear to be tied to the stimulus
presentation, these should not be coded as facial expressions. Determination of this can be
difficult and is best established by observing several occurrences of such movements across
stimulus presentations.
Gum chewing
Participants chewing gum can present a sticky problem for coders. Gum chewing may
actually inhibit natural expressive displays. The best solution here is to make sure a participant
removes gum prior to the beginning of stimulus presentation.
Talking
Talking during a study can be problematic if more than one participant is being run
through a study or if an experimenter is in the room. The best advice is to strongly encourage
participants to refrain from talking during the experiment. In the event that coders are faced
with rating a segment in which a participant is talking, attempts should be made to identify an
expression independent of the talking. For example, if a participant smiles and then begins
talking, the smile should be recorded as an expression. If on the other hand, the participant
begins talking and has clearly diverted attention from the stimulus presentation, and smiles, it
should not be recorded as an expression.
Facial tics
Occasionally, a participant may repeatedly display facial movements that do not appear
to be expressions of emotion and are instead facial tics. As cited above, contact lens wearers
7
may have eye movements that are related only to the lenses. Other people may have other
repetitive facial movements. These may not be obvious initially, but after viewing several
minutes of a participant's record, they may become more prominent. A special case involves
psychiatric patients with tardive dyskinesia. Patients who have taken neuroleptic medication
for long periods of time may develop this very unfortunate side effect. Tardive dyskinesia
involves uncontrollable repetitive movements that may involve facial muscles, most often those
around the mouth. Any work done with psychiatric patients should involve careful assessment
of these symptoms.
Since its inception, reliability for coders using FACES has been calculated using the
intraclass correlation coefficient. The intraclass correlation coefficient is the correlation
between one measurement (e.g., ratings of facial expressions) on a target and another
measurement made on that same target (Shrout & Fleiss, 1979). More specifically, following
the Case 2 study described by Shrout & Fleiss (1979), the coders (judges) are considered to be
selected from a random sample of judges, and each judge rates each subject or target. That is, it
is assumed that FACES can be used effectively by any set of coders. The formula used to
calculate the ICC is derived from the components of a two-way ANOVA (Subjects x Coders)
which partitions the within-target sum of squares into a between-coders sum of squares and a
residual sum of squares. Because the variance to due coders is not ignored, the coefficient can
be interpreted as an index of agreement rather than consistency (Shrout & Fleiss, 1979). As
such, the formula is:
where:
BMS = between subjects mean square
EMS = residual mean square
CMS = between coders mean square
k = number of coders
n = number of subjects
In our applications, using trained undergraduate and graduate students as coders and
with varied participant populations (e.g., undergraduates, adult community residents,
psychiatric patients), the agreement has been very high, ranging from .70 to .99.
8
References
Adelmann, P. K., & Zajonc, R. B. (1989). Facial efference and the experience of emotion.
Annual Review of Psychology, 40, 249-280.
Darwin, C. R. (1896). The expression of emotions in man and animals. New York: Appleton.
Ekman, P. (1982). Methods for measuring facial action. In K. R. Scherer & P. Ekman (Eds.),
Handbook of methods in nonverbal behavior research. Cambridge: Cambridge University
Press.
Ekman, P., & Friesen, W. V. (1976). Measuring facial movement. Journal of Environmental
Psychology, 1, 56-75.
Ekman, P., & Friesen, W. V. (1978). The Facial Action Coding System. Palo Alto, CA:
Consulting Psychological Press.
Ermiane, R. & Gergerian, E. (1978). Atlas of Facial Expressions. Album des expressions du
visage. Paris: La Pensee Universelle.
Gottman, J. M., & Levenson, R. W. (1985). Assessing the role of emotion in marriage.
Behavioral Assessment, 8, 31-48.
Izard, C. (1979). The maximally descriminative facial movement coding system. (MAX).
Unpublished manuscript. Available from Instructional Resources Center, University of
Delaware, Newark.
Larsen, R. J., & Diener, E. (1992). Promises and problems with the circumplex model of
emotion. In Margaret Clark (Ed.), Review of Personality and Social Psychology, 13, 25-
59.
Notarious, C., & Levenson, R. (1979). Expressive tendencies and physiological response to
stress. Journal of Personality and Social Psychology, 37, 1204-1210.
9
Shrout, P. E. & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin, 86, 420-428.
Tomkins, S. S. (1962). Affect, Imagery, Consciousness: Vol 1. The Positive Affects. New York:
Springer.
Tomkins, S. S. (1963). Affect, Imagery, Consciousness: Vol 2. The Negative Affects. New York:
Springer.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief
measures of positive and negative affect: The PANAS scales. Journal of Personality
and Social Psychology, 54, 1063-1070.
10
Appendix
11
Positive and Negative Affect Descriptors
Positive Negative
Happy Miserable
Delighted Distressed
Glad Annoyed
Amused Jittery
Pleased Nervous
Content Angry
Satisfied Gloomy
Calm Anxious
Serene Afraid
Excited Tense
Astonished Alarmed
Cheerful Frustrated
Surprised Disgusted
Active Depressed
Content Hostile
12
LIKERT FORMAT SUMMARY SHEET:
Please rate the degree to which the participant expressed each of the following emotions using the scale below:
What is the overall level of expressiveness for this person for this film?
(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(;(
Please rate the degree to which the participant expressed each of the following emotions using the scale below:
What is the overall level of expressiveness for this person for this film?
13
FORCED CHOICE FORMAT SUMMARY SHEET
SUMMARY SHEET
What was the predominant emotion being expressed throughout the clip?
A. Interest
B. Sadness
C. Anger
D. Surprise
E. Fear/Disgust
F. Neutral/Indifferent
G. Happiness
What is the overall level of expressiveness for this person for this clip?
1 2 3 4 5
14